Model: GELU Model: 4 Layers, 2048 Neurons per Layer

Dataset: 80% C4 (Web Text) and 20% Python Code

Neuron 1996 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.7153. Min Act: -0.1700

Data Index: 50424 (C4 (Web Text))

Max Activating Token Index: 1013

Text #1

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.6527. Min Act: -0.1700

Data Index: 403199 (C4 (Web Text))

Max Activating Token Index: 966

Text #2

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.5588. Min Act: -0.1700

Data Index: 1081378 (C4 (Web Text))

Max Activating Token Index: 595

Text #3

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.5275. Min Act: -0.1700

Data Index: 1565825 (Python Code)

Max Activating Token Index: 581

Text #4

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.5275. Min Act: -0.1700

Data Index: 1425768 (Python Code)

Max Activating Token Index: 689

Text #5

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.5588. Min Act: -0.1700

Data Index: 1610202 (Python Code)

Max Activating Token Index: 269

Text #6

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.5275. Min Act: -0.1700

Data Index: 1251085 (C4 (Web Text))

Max Activating Token Index: 566

Text #7

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.5275. Min Act: -0.1700

Data Index: 167899 (C4 (Web Text))

Max Activating Token Index: 464

Text #8

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4961. Min Act: -0.1700

Data Index: 1358713 (C4 (Web Text))

Max Activating Token Index: 770

Text #9

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4961. Min Act: -0.1700

Data Index: 1420633 (Python Code)

Max Activating Token Index: 489

Text #10

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4961. Min Act: -0.1700

Data Index: 1574505 (Python Code)

Max Activating Token Index: 57

Text #11

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4648. Min Act: -0.1700

Data Index: 694825 (C4 (Web Text))

Max Activating Token Index: 749

Text #12

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.3393. Min Act: -0.1700

Data Index: 488673 (C4 (Web Text))

Max Activating Token Index: 445

Text #13

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4334. Min Act: -0.1700

Data Index: 1503002 (Python Code)

Max Activating Token Index: 670

Text #14

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4334. Min Act: -0.1700

Data Index: 1399510 (Python Code)

Max Activating Token Index: 674

Text #15

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4334. Min Act: -0.1700

Data Index: 1362624 (Python Code)

Max Activating Token Index: 690

Text #16

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4021. Min Act: -0.1700

Data Index: 1456140 (Python Code)

Max Activating Token Index: 868

Text #17

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4334. Min Act: -0.1700

Data Index: 1582911 (Python Code)

Max Activating Token Index: 641

Text #18

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4334. Min Act: -0.1700

Data Index: 64422 (C4 (Web Text))

Max Activating Token Index: 756

Text #19

Max Range: 3.7153. Min Range: -3.7153

Max Act: 3.4648. Min Act: -0.1700

Data Index: 824196 (C4 (Web Text))

Max Activating Token Index: 409